Hi Mark, 

maybe the first sentence was a wrong entry and for the problem not relevant. 
The important issues are:

> We can elect between 2 problems:
> A) Performance problem with the 1.2.18 
> B) Problem in file uploads with 2.0.4 in 64 bit environment

The installation of the connectors are ok, the installation of the tomcat is 
fine but the performance of the 1.2.18 connector is wrong in comparison with 
the 2.0.4 connector version. 
And here, we thought, maybe exist some adjusting screws inside of the code to 
get a throughput like the 2.0.4 version.
We think, we don't find a solution for this in the user list. 

Enno

P.S.Thank you for the quick answer.



-----Ursprüngliche Nachricht-----
Von: Mark Thomas [mailto:[EMAIL PROTECTED] 
Gesendet: Montag, 7. August 2006 13:29
An: Tomcat Developers List
Betreff: Re: Jakarta Connectors - a) Performance Differences or b) Problems 
with 64bit version

Lichtenfeld, Enno, ext. wrote:
> Hi,
> 
> we need a Jakarta connector for Win2K3 Server (x64) with IIS 6.0.

This question belongs on the users list since it is about how to use Tomcat.

The dev list is for questions related to the development (ie changes in source 
code) of Tomcat.

Mark

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: 
[EMAIL PROTECTED]
==========================================================================


Hi,

we need a Jakarta connector for Win2K3 Server (x64) with IIS 6.0.

Till now we used the connector 2.0.4, because this connector contained features 
we used, and in 1.2.x we missed this features. 
With features of the newest versions of 1.2 we can now also use version 1.2.18. 
And because the version 2.0.4 is deprecated we want to use in future the 
version 1.2.18.

We can elect between 2 problems:
A) Performance problem with the 1.2.18 
B) Problem in file uploads with 2.0.4 in 64 bit environment

What can we do to get a right functional connector for us?


Topic A
=======
Versions used for comparative test:

2.0.4:  For the 64bit compatibility we compiled the sources with 
        MS Visual Studio 2005 and got a 64Bit (x64) dll.
1.2.18: We use 
        - the downloaded binary or 
        - an own build version (also Visual Studio 2005) with different 
optimizations.

Test case:
"Web Application Stress Tool" with a permanent click stream of 90 clients 
against the jsp-examples.

Test environment:

2 64Bit Server Win2K3
Machine One: Web Application Stress Tool, IIS 6.0, and Jakarta Connector 
Machine Two: Tomcat 5.0.28 with jsp-examples

Test duration: 3 Minutes

The Load whiles the test:
Machine One: in both cases 100%
Machine Two: - with Jakarta Connector 1.2.18: ~4%
             - with Jakarta Connector 2.0.4: ~40%

Result is nearly independent of optimization inside the VS 2K5. We used no 
debug versions.

AND NOW:

Throughput:
Jakarta Connector 1.2.18: 270000 Hits and 120 NA 
Jakarta Connector 2.0.4:  400000 Hits (All values roundabout values.)

Remark: The performance difference between both connectors is the same, if "Web 
Application Stress Tool" runs on the second machine. 

In Summary:
- the performance is much better in 2.0.4 
or
- we do not know all adjusting screws in 1.2.18. Maybe there are some 
possibilities for tuning. 

Is this explainable? 


____________________________________________________________________


Topic B
=======

On a web application (struts inside) with a file upload page, tomcat receives 
files, which are different to the file before upload. We saw it also with a 
network stream scanner between IIS and Tomcat. Every x thousand bytes something 
adds some bytes in the upload stream. 

This behaviour occurs only under the following conditions:
- The web access happens over a WAN (slow network?)
- The connector runs not in DEBUG mode
- The environment is a 64Bit environment

Do you know this behaviour?


____________________________________________________________________


Can somebody give us a help? 

Many Thanks.










---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to